Research On Topic-based Influence Maximization In Academic Network

As the important form of scientific research results,academic literature can form a network through mutual citations,providing a wealth of reference materials for scientific researchers.A key problem around the application of the academic literature network is how to identify the most important papers,this problem can be modeled as the influence maximization problem in the academic network.Although the influence maximization in other fields has received extensive attention from the research community,most of the existing research results ignore the topic factor and are not suitable for the academic network.Thus,in terms of topic-based influence maximization problem in the academic network,this paper constructs an influence propagation model based on the topic distribution,and further proposes the most influential node acquisition algorithm based on topic in the universal set and the specify target set.Main research contents of this paper are as follows:(1)Influence propagation model based on topic distribution.At present,most influence propagation models ignore the topic factor when setting the activation probability between nodes.However,the topic distribution of the information on nodes in the academic network plays an important role in calculating activation probability between nodes.To better simulate the process of influence propagation in academic networks,this paper builds an independent cascade model based on topic distribution after improving the traditional independent cascade model.Experiments show that the model is feasible in solving the problem of influence propagation in academic networks.(2)The most influential node acquisition algorithm based on topic in the universal set.Algorithms to solve the problem of influence maximization can be mainly divided into two types: greedy strategy algorithm and heuristic algorithm.Among them,the algorithm based on greedy strategy has higher accuracy,but lower speed;while the heuristic algorithm runs faster,but the influence range of the algorithm cannot be guaranteed.So this paper proposes a topic-based mixed type algorithm in the academic network combining with the advantages of heuristic algorithm and greedy strategy algorithm.Experiments show that the influence range of the algorithm proposed in this paper is very close to the greedy strategy algorithm,and the running time is reduced by 36.2% compared with the greedy strategy algorithm.(3)The most influential node acquisition algorithm based on topic in the specify target set.The traditional influence maximization problem is to find the most influential nodes in the entire network,while local influence maximization problem is to specify a target node and find a seed set to maximize the influence of the target node.As the current local influence maximization algorithm only considers the network topology and local attributes,this paper proposes a topic-based hop algorithm in the academic network.Experiments show that the algorithm proposed in this paper has an influence of 11.3% higher than the LND algorithm on the target node.